Abstract: The vibration and noise
signals are detected for classifying and discriminating the
states of tool dulling also for gearing wearing and forecasting
tool wearing. The correct discrimination of tool dulling is
increased by using boundary samples as training sets. A linear
discriminatory structure of twice dichotomizer is put forward.
Excellent results have been obtained for discriminating three
nonuniform states of gearing wearing. Forecasting accuracy is
increased by correcting forecast.
Key words:
Discriminate Autoregressive model Optimal forecast
Manuscript received on March
15, 1993
|